servo vs one shot

J

joescrivens

Suspended / Banned
Messages
15,052
Name
Joe
Edit My Images
Yes
I've been shooting AI Servo for about a year now even if the subject is static and today for some reason I decided to just switch to one shot and see if i changed my mind

The result is that servo seems much less reliable than one shot for static objects. AI Servo is performing really badly in comparison look:

one shot - focussed on the O

IMG_0524 by MrJoeBoy, on Flickr

crop2 by MrJoeBoy, on Flickr


servo - focussed on the O

IMG_0522 by MrJoeBoy, on Flickr

crop4 by MrJoeBoy, on Flickr

these aren't isolated results, i am constantly getting better focus of static objects with one shot with all different lenses.



Is this normal?
 
Last edited:
Have you got an intelligent focus mode ie it is 'one shot' for something static and continuous for something moving? I'd never try continuous for anything static as I don't want the camera buggering about. I usually lock focus with one shot then recompose. Servo will be constantly seeking out the areas of highest contrast which is why it's drifted off towards the edge of the box. Only takes a tiny twitch to make it want to drift off too.
 
AI Servo isn't really meant to be used for Static objects.

It only really works well with moving objects, ie, football players, cars etc.

Why would you want to use AI Servo on static objects anyway?

But to your original question, it looks like you were using multiple focus points instead of just one single focus point, which is what you'd want if you were to use Servo on static for some bizzare reason...

:)
 
Last edited:
Joe, I agree with you. There are times when simulating one shot mode (with back button AF) while really in servo mode won't cut it. In particular that's when you have very shallow DOF or extremely poor light/contrast. On those occasions, so long as neither subject nor camera is moving I do think one shot gives more consistent results. I don't think anybody would recommend using AI Servo when performing AF calibration, for example.

One factor in that, especially in the low light scenario, is that the camera takes a little more time to collect AF data before deciding what to do when in one shot mode. In servo mode, with the constant time pressure to be recalculating where focus should be, the camera samples for a shorter period of time before deciding where to send the lens. That information was published by Chuck Westfall in one of his missives a while back. It makes sense to me.

With that said, my cameras spend most of their lives in servo AF mode, regardless of subject matter, but if I spot difficulties and realise that one shot will help then I will switch modes as necessary to achieve the best results. On my 7D I have set the DOF button to temporarily swap AF modes, so I can usually leave the camera in servo mode but quickly override to one shot mode without faffing about too much. One example where I might choose to shoot like that is when shooting birds, mostly in flight, but then I might choose a floater for a shot or two and temporarily switch over to one shot mode before carrying on with my BIF.
 
inlineadam said:
AI Servo isn't really meant to be used for Static objects.

It only really works well with moving objects, ie, football players, cars etc.

Why would you want to use AI Servo on static objects anyway?

But to your original question, it looks like you were using multiple focus points instead of just one single focus point, which is what you'd want if you were to use Servo on static for some bizzare reason...

:)
Click to expand...

It's not that I WANT to use servo for static objects it's that most of the time my objects move and when they stay still for the brief second I don't change modes. I wasn't using multiple point focus I was using single centre point for both using back button focus
 
Last edited:
tdodd said:
Joe, I agree with you. There are times when simulating one shot mode (with back button AF) while really in servo mode won't cut it. In particular that's when you have very shallow DOF or extremely poor light/contrast. On those occasions, so long as neither subject nor camera is moving I do think one shot gives more consistent results. I don't think anybody would recommend using AI Servo when performing AF calibration, for example.

One factor in that, especially in the low light scenario, is that the camera takes a little more time to collect AF data before deciding what to do when in one shot mode. In servo mode, with the constant time pressure to be recalculating where focus should be, the camera samples for a shorter period of time before deciding where to send the lens. That information was published by Chuck Westfall in one of his missives a while back. It makes sense to me.

With that said, my cameras spend most of their lives in servo AF mode, regardless of subject matter, but if I spot difficulties and realise that one shot will help then I will switch modes as necessary to achieve the best results. On my 7D I have set the DOF button to temporarily swap AF modes, so I can usually leave the camera in servo mode but quickly override to one shot mode without faffing about too much. One example where I might choose to shoot like that is when shooting birds, mostly in flight, but then I might choose a floater for a shot or two and temporarily switch over to one shot mode before carrying on with my BIF.
Click to expand...


Cheers Tim.

It's never stopped me getting focus at 1.4 before but maybe those missed shots I've had this year were due to this issue?
 
But what you've seen is possible right. In AF servo the subject is continuously focussed. Depending on your AF point selection, if the subject moves away from the focus point, then focus tracking continues as soon as another focus point grabs it.

Doesn't it also start with predictive focussing?

Interesting article here: http://photonotes.org/other/ai-servo.html
 
Byker28i said:
But what you've seen is possible right. In AF servo the subject is continuously focussed. Depending on your AF point selection, if the subject moves away from the focus point, then focus tracking continues as soon as another focus point grabs it.
Click to expand...

no.

When I select the centre point and choose servo it continuously focuses whatever is on the centre point, if I focus on my finger then move my finger to the right hand focus point then it refocuses to whatever is on the centre point.

It doesn't track the subject across the focus points.
 
joescrivens said:
no.

When I select the centre point and choose servo it continuously focuses whatever is on the centre point, if I focus on my finger then move my finger to the right hand focus point then it refocuses to whatever is on the centre point.

It doesn't track the subject across the focus points.
Click to expand...

Regarding the focus question in the examples..

If you were only using the centre point, then I would have thought that the system is still expecting or trying to predict movement and is therefore probably 'hunting' by a small amount due to you hand holding the camera.
 
Last edited:
joescrivens said:
no.

When I select the centre point and choose servo it continuously focuses whatever is on the centre point, if I focus on my finger then move my finger to the right hand focus point then it refocuses to whatever is on the centre point.

It doesn't track the subject across the focus points.
Click to expand...

Yup, that's right. I did say it depends on your focus point selection.

Based on it's predictive design, is it hunting slightly? That might explain it

Interesting this. I tend to use single shot mostly on static subjects, only switching to servo for sport. Never thought about using it any other way, probably because of the way the manual was written.
 
The results are qualitatively as I would expect, but I am surprised by the extent.
 
it's very strange indeed to see results so dramatic, i've been using servo for all my focussing for over a year. My children are moving all the time so it's easier for me to keep it in servo but when they are still I still use it and have never seen results this poor and I always shoot wide open on fast primes

this shot was done in servo


IMG_0281 by MrJoeBoy, on Flickr

and there were no focussing problems as you can see
 
Did you use a spot AF point for the shot in the first post or a standard single point? The spot AF point is supposedly the size of the AF point markers in the viewfinder and the standard AF point is a bit larger than the marked box. When you aim at an angled target there is no guarantee which part of the target the camera will end up focusing on. That's why you should not use an angled target when calibrating AF. It's possible that in servo mode the camera chose to focus on a slightly different position and the servo quirkery might have caused the focus to drift a little further.

If you have DPP installed then you can use it to show the exact AF point placement and the size of the indicated AF point within the frame. A standard AF point (not spot) might find something to focus on outside the marked boundary of the AF point.
 
If you press CTRL-I it will bring up the info panel and (I think) that should tell you the AF point mode selected. If you had a standard AF point rather than spot I can definitely see the possibility that the AF might have been drawn forwards towards the "R".

Doh! you're using a 5D2, not a 7D. That means you will only have standard AF points available, which means they will indeed be larger than the squares indicate. That means you cannot guarantee with absolute certainty where the AF will end up when aimed at an angled target with contrasty areas in several places. Odds are that it will go for the closest point (more or less), but not guaranteed.
 
Last edited:
tdodd said:
If you press CTRL-I it will bring up the info panel and (I think) that should tell you the AF point mode selected. If you had a standard AF point rather than spot I can definitely see the possibility that the AF might have been drawn forwards towards the "R".

Doh! you're using a 5D2, not a 7D. That means you will only have standard AF points available, which means they will indeed be larger than the squares indicate. That means you cannot guarantee with absolute certainty where the AF will end up when aimed at an angled target with contrasty areas in several places. Odds are that it will go for the closest point (more or less), but not guaranteed.
Click to expand...

i see, so in your opinion it's all about this angled problem with a larger af point?
 
joescrivens said:
i see, so in your opinion it's all about this angled problem with a larger af point?
Click to expand...

Largely, perhaps. I mean, when you aim at an angled target and you are shooting with shallow DOF where do you think the focus should end up? The camera can only focus at one distance, but you are presenting it with a target, within the AF sensor's scope, which is at various distances. It can only pick one. Which one it chooses....... a bit of a lottery maybe.

This is why you do not try to calibrate AF by shooting at angled rulers. There are far too many possibilities for where the camera might end up focusing. That's why the best target for checking AF accuracy is something along the lines of LensAlign - http://www.whibalhost.com/la/pressroom/.
 
Last edited:
tdodd said:
On my 7D I have set the DOF button to temporarily swap AF modes, so I can usually leave the camera in servo mode but quickly override to one shot mode without faffing about too much.
Click to expand...

:thumbs::thumbs::thumbs:

Rushes off to consult 7D manual with camera in hand :D

David
 
Agree with Tim. Never use an angled target for AF checks.

And the framing of the two comparison shots is different. Second pic camera has moved left so it's picked up on the R.
 
HoppyUK said:
Agree with Tim. Never use an angled target for AF checks.

And the framing of the two comparison shots is different. Second pic camera has moved left so it's picked up on the R.
Click to expand...

+1 to Tim and Hoppyuk
 
You must log in or register to reply here.
Back
Top